KEYWORDS
Natural language processing, machine learning, supervised learning, deep learning, neural networks, word embeddings, recurrent neural networks, sequence to sequence models
Artificial Neural Networks or Connectionist--------->>
- To Develop a Machine that is capable of thinking like a human
- Neural Means: What are the Neurons
- Network Means: How they are Connected
- Neuron Means Thing that holds a number ------>>> eg 0.2 or 05 etc.
- Neural Network: Input Layer + Hidden + Output + weights and Biases + Activation Function means sigmoid
Calculating the predicted output ลท, know as Feedforward
Updating the weights and biases, know as backpropagation
Tool Use Machine Learning: Anaconda Python
- Machine Learning: Technology + Engineering + Algorithms + Data + Network + Computer science + Artifical Intelligence.
- Connectionist System + Biological Neural Network
- ANN is not algorithm + Many different Machine Learning Algorithm work together and process complex date inputs.
- Any task-specific rules & regulation of (ANN)
- ANN collection of connected units or nodes called artificial neurons.
- Each connection, like the synapses in a biological brain, can transmit information, a 'single' form one artificial neuron to another.
- An interconnected group of nodes + Brain of network + Each circle node represents Artificial Neuron + Arrow is connection + Input and Output, Hidden Layer
- Sum of Inputs + Connection called edges + Weights ----> increase and decrease straight of single at the connection.
- Single travel first layer to the last layer means the output
- ANN used variety fo tasks including computer vision, speech recognition, machine translation, social network, filtering, playing board, and video games and medical diagnoses.
Chapter 1
- Def Machine Learning = Algorithm + Statistical Models Analysis + input + output
- Training Data ------> Mathematical Model + Maths Calculation + Study of construction of alg + Multiple datsets results + validation dataset + Hyper-parameter + Regulationzation + Over-Fitting, Under Fitting + ill-posed prblem + Test dataset + holdout datset .
- Machine Learning Tasks: Supervised Learning + Semi-Supervised Learning + Classification Algorithm + Unsupervised Algorithm.
- Active Learning Algorithm + Reinforcement Learning Algorithms, + Meta-Learning Algorithm + Robot Learning Alg.
- Machine Learning: Process and Techniques ------->> Feature Learning + Sparse dictionary learning + Anomaly detection + Decision trees + association rules.
- Machine Learning Models: ANN + Deep Learning + Support Vector Machines + Bayesian Network + Genetic Algorithm
- Evolutionary Algorithms: Evolutionary computation + Meta-Heuristic Algorithm + optimization alg + stochastic optimization + combinatrial optimization + Swarm Alg + Ant colony otimization + Artifical bee colony alg + Cuckoo Search alg + Particle Swar optimization
- Meta- Heuristic Alg + Hunting search + Adaptive dimensiaonl search + Firefly alg + Harmoney search + Guassion adaptation + memetic alg + Emperor penguines colongy
- Biase-Variance Tradeoff: Function complexity and amount of training date + Dimensionality fo the input space + Noise in the output
- Bais + Forward Propagation & Back Propagation + Activation Function -----> Sigmoid Function
- PSO Algorithm + Bat Algorithm + Grey Wolf Algorithm + Dolphin Algorithm + Genetic Alg + Hinting Alg +
- What is Intelligence? What is AI? What is Machine L
It can be described as the ability to gather information and to retain it as knowledge to be applied in some situation. Intelligence comes from the capacity of logic + Understand the things + self-awareness + Learning + Emotional Knowledge + Reasoning + Planning + Creativity + Problem Solving.
- The simple words, it is the capability of a machine to mimic intelligent human behavior + Weak and strong AI
What does a machine need to be an Intelligent
1. Perception: Understanding Images, Audio, etc.
2. Reasoning: Answering questions from data
3. Planning: Inferring the required steps to reach a goal
4. Natural Language Procession - Understanding human language
The Biggest Challenges Facing Artifical Intelligence
- Create Model Learn Faster
- Accurate Response + Recurrent Connections + Memory Storages
- Convert Speech to Text form + AI Identify this is a pen thin or fit, dog, cat, school
- Statistics reveal that 55% of survey respondents felt the biggest challenges was the changing score of human jobs when everything will be automated.
Optimization ALgorithms:
Some Examples of optimization algorithms include:
- ADADELTA + ADAGRAD + ADAM + NESTEROVS + NONE + RMSPROP + SGD + CONJUGATE GRADIENT + HESSIAN FREE + LBFGS + LINE GRADIENT DESCENT
ACTIVATION FUNCTIONS
The activation function determines the output a node will generate, based upon its input. The activation function is set at the layer level and applies to all neurons in that layer.
- CUBE + ELU + HARDSIGMOID + HARDTANH + IDENTITY + LEAKYRELU + RATIONALTANH + REULU RRELU + SIGMOID + SOFTMAX + SOFTPLUS + SOFTSIGN + TANH
The Machine Learning Algorithm List Includes:
1. Linear Regression 2. Logistic Regression 3. Support Vector Machines 4. Random Forest 5. Naive Bayes Classification 6. Ordinary Least Square Regression 7. K-means 8. Ensemble Methods 9. Apriori Algorithm 10. Principal Component Analysis 11. Singular Value Decomposition 12. Reinforcement or Semi-Supervised Machine Learning 13. Independent Component Analysis
Deep Neural Networks for Text Classification
The Challenges of Natural Language Processing?
- NLP is the field of Design methods & Algorithms to take input and generate output.
- Challenge 1: (consider the sentence I ate pizza with friends, and compare it to I ate pizza with olives)
- Set of rules for challenging tasks & readers can easily categorize a document into its topic, Language is symbolic and discrete.
- words—there is no simple operation that will allow us to move from the word “red” to the word “pink” color of words. A phrase can be larger than the meaning of the individual words.
Classification: Words + Sentence + Phrase + Video + Image?
1 Text Classification
Breast Cancer Research Tasks?
- Site Using in Research: 1 Sci-Hub 2 Kaggle 3 Dataset Research 4 Google Scholar
Breast Cancer
- Describe quantitive mass-spectrometry
- Based on Proteomic and phosphoproteomic Analysis of 105 genomically annotated breast cancer.
- 77 Higi Quality data + Chromosomal Loss --------- Make DNA and Protein
- SQ trans + CETN2 and SkIP
- What are Proteins + Clusters and Pathway analysis + Identify mRNA Level + Other also associated to identify
- Genome in cancer + what is Tumor? + Four principles MRNA- define breast cancer intrinsic subtypes. + Analysis of 105 breast tumor.
- Total Proteins 15,369 + 12,405 Genes and Phosphosites 62,679 + 11,632 Proteins per tumor. + Total 90,806 + DNA 84,667 + RNA 54,201 Somatic variants + proteome variants
- Direct Effects of genomic alternations on Proteins level?
RNA and DNA + Protein Expression of breast-cancer relevant genses across tumor + Frequently detected and differential phosphorite ration show for each gene.
We observed at the peptide level by searching MS / MS Spectra + Database + Amino acid level with current technologies
- Proteins: Give Food to child + Protein make a body of human + Amnesty ----- Structure ---- 1 Corbaorail Group 2 Hydron 3 Amion group 4 Alcoyal Group
- Peptide: Bands b/w proteins called peptide
- Spectrometric: Measure radiation + - Proteome: Hydrogen Isotop, Cellular ---- means multiple cells
- Composition and Structure + Genes + Living Cell + Protegenomics ----> is a field of biology.
- Tumors: When cells become increase then made tumors.
-What is DNA and RNA: DNA Means Deoxyribonucleic Acid OR RNA Means Ribounucle Acid
DNA --- Replicates, and stores genetic information. It is a blueprint for all genetic information contained within an organism.
RNA: Converts the genetic information contained within DNA to a format used to build proteins, and then moves it to ribosomal protein factories.
- What's New in Breast Cancer Research?
1 Breast Cancer Causes 2 Causes and treatment of metastatic breast cancer 3 Reducing breast cancer risks
- Scientists trained a computer to classify breast cancer tumors
288 Images to test a computer's ability to distinguish features of the tumor + Computer test sperate genes
- Classification Based on Breast cancer types
1 Classification based on breast cancer stage 2 Classification based on breast gene Expression 3 Classification based on tumor
#Train/Test Split and Cross Validation in Python
# linear regression in Python
#Pandas — to load the data file as a Pandas data frame and analyze the data.
#From Sklearn, I’ve imported the datasets module, so I can load a sample dataset,
#and the linear_model, so I can run a linear regression
#From Matplotlib I’ve imported pyplot in order to plot graphs of the data
